DinoLink: A Token-Centric Representation Compression Framework for Bandwidth-Constrained Collaborative V2X Perception

2026-06-24

AI Summary

This work addresses the challenge of high-precision remote perception in vehicle-to-everything (V2X) scenarios, where extremely limited bandwidth hinders efficient transmission of raw perceptual data. To overcome this, the authors propose a token-centric semantic communication framework featuring a novel dual-sparsity architecture: a saliency-aware token selector discards redundant background information, while residual vector quantization (RVQ) compresses features into compact codebook indices. Only lightweight indices and positional priors are transmitted, departing fundamentally from conventional pixel-stream paradigms. Evaluated on the nuScenes dataset, the method achieves a 139× compression ratio while retaining a 32.8% mAP, and demonstrates a 34.5× speedup in inference under narrowband conditions such as LoRa.

Abstract

High-precision remote perception is often hindered by the severe bandwidth constraints of Vehicle-to-Everything (V2X) networks. We propose \textit{DinoLink}, a token-centric compression framework that replaces raw pixel streaming with discrete semantic communication for vehicle-cloud collaborative inference. DinoLink employs a dual-sparsity architecture: a saliency-aware selector prunes redundant background tokens, while a Residual Vector Quantization (RVQ) module collapses features into compact codebook indices. By transmitting only lightweight indices and positional priors, DinoLink achieves a $139\times$ bitrate reduction compared to uncompressed transmission while maintaining a competitive 32.8\% mAP on the nuScenes dataset. Deployment simulations further demonstrate a $34.5\times$ acceleration in narrow-band environments, such as LoRa. Our results substantiate DinoLink as a robust, bandwidth-efficient frontend for high-fidelity remote perception in constrained V2X scenarios. The code is publicly available at https://github.com/UGA-MOBILITY-LAB/dino_link.

Problem

• V2X perception
• bandwidth constraint
• remote perception
• representation compression
• collaborative inference

Innovation

• token-centric compression
• semantic communication
• Residual Vector Quantization (RVQ)
• bandwidth-constrained V2X
• collaborative perception